Computer vision
NLP
Tabular Data
Collaborative filtering
Recommendation engine
Embeddings
resnet34 and resnet50 are the type of architectures in CNN (convolutional neural network)
every dataset in fast.ai has x and y
ipywidget is a UI library in python - check it out
Inference is the prediction part of machine learning. When the model is already in production and you send a data for prediction to the model.
You have got a good learning rate WHEN
<aside> 💡 increasing it increases the losses
</aside>
<aside> 💡 decreasing it - slows down the learning
</aside>
too few epochs
training error is much leser than validation error. Keep an eye on these two errors, the difference must not be that much